Visual animals utilize spectral information in two ways: true color vision, which differentiates spectral compositions largely independent of brightness, allows animals to recognize objects and register and retrieve color memory; innate spectral preference, which depends on intensity but not learning, often reflects individual species' specific ecological needs. Using a combination of genetic, histological, electrophysiological, imaging and behavioral approaches, our group studies how visual circuits process chromatic information to guide behaviors in Drosophila. Our strategy is (i) to identify key neuronal types and to map their synaptic connections, (ii) to examine functional requirement of identified neurons for color vision and spectral preference behaviors, and (iii) to determine the synaptic mechanism that transform visual signals at different processing stages. Using molecular genetics and histology and serial-EM reconstruction, we mapped the synaptic circuits of the chromatic photoreceptors, R7s and R8s, and their synaptic target neurons, Tm and Dm neurons, in the peripheral visual system as well as their downstream targets in the higher visual system, the lobula. The medulla projection (Tm) neurons (Tm5a/b/c, Tm9 and Tm20), analogous to vertebrate retinal ganglion cells, relay photoreceptors to higher visual centers while the Dm (Dm8) neurons provide an indirect pathway by relaying photoreceptors to Tm neurons. To probe the synaptic connections between these neurons, we developed several versions of GRASP (GFP reconstitution across synaptic partners) methods to map active synapses. Using these methods, we found that the chromatic photoreceptors, R7 (UV-sensing) and R8 (blue/green-sensing), provide inputs to a subset of first-order interneurons. Tm9/20/5c and Tm5a/b receive direct synaptic inputs from retinotopic R8s and R7s, respectively, consistent with their functions in processing single visual pixel information. In contrast, the amacrine neuron Dm8 pools inputs from 14 R7s and provides input for Tm5c. Based on histology, we found the Tm5a/b/c and Tm20 neurons relay photoreceptor signals to the higher visual center, the lobula, which in insects has been implicated in processing and relaying color information to the central brain. Using the GRASP method and a series of lobula neuron drivers, we identified the lobula neurons that receive direct synaptic inputs from the chromatic Tm neurons, Tm5a/b/c and Tm20. We found four types of lobula neurons that form synaptic contacts with chromatic Tm neurons: two novel lobula intrinsic neurons, Li3 and Li4, and two lobula projection neurons, LT11 and LC14. Each LT11 elaborates a large dendritic tree to cover the entire lobula layers, Lo4-6, and projects an axon to optic glomeruli in the central brain. Each Li4 extends dendrites to cover about 60% of the lobula layer, Lo5 while each Li3s dendrites cover about 15% of the lobular layers 5 and 6. To confirm synaptic connections observed at the light microscopic level, we developed a two-tag EM double labeling technique that highlights both presynaptic and postsynaptic terminals in the same preparation for EM analysis. By combining two orthogonal expression systems and two different peroxidases, HRP (horse radish peroxidase) and APX (ascorbate peroxidase), we expressed HA-tagged mitochondrion-targeted APX in Tm5c and membrane-tethered HRP in their postsynaptic neurons LT11 and confirmed appositions among synaptic profiles. Our anatomical study revealed that the lobula neurons integrate multiple chromatic inputs from Tm neurons over a large receptive field, suggesting that color vision is of low spatial resolution. To identify neurons that is involved in innate spectral preference, we used genetic methods to inactivate systematically different first-order interneurons and examined their behavioral consequences. We previously found that the amacrine Dm8 neurons, which receive UV-sensing R7 photoreceptor inputs, are both required and sufficient for animals' innate spectral preference to UV light. Using behavioral assays, we further found that inactivating Tm5c, one of Dm8s synaptic targets, abolished UV preference, establishing Tm5c as the key downstream targets for spectral preference. Using single-cell transcript profiling and immunohistochemistry methods, we found that both Dm8 and Tm5c express vesicular glutamate transporter (VGlut) and that Tm5c expresses four kainite-type glutamate receptors (Clumsy, DKaiR1C, DKaiR1C, CG11155). RNAi-knockdown of these receptors in Tm5c, thus preventing Tm5c from receiving Dm8 inputs, abolished UV preference, indicating that these receptors are functionally required in Tm5c for UV preference behaviors. Thus, the R7s->Dm8->Tm5c connections constitute a hard-wired glutamatergic circuit for detecting dim UV light. To determine the functional properties of these CNS glutamate receptors, we reconstituted DKaiR1D in oocytes and determined its electrophysiological properties. We found that DKaiR1D forms a rapidly activating and desensitizing homotetramer receptor that is glutamate-gated and calcium permeable. Notably, DKaiR1D is inhibited by NMDA and AP5, the agonist and antagonist of vertebrate NMDA receptors, indicating that insect glutamate channels have very different pharmacology from vertebrates. Two-photon calcium imaging of Dm8 and Tm5c expressing the calcium indicator GCaMP6f revealed that the activity of both neurons is suppressed by full-field UV light illumination. The light responses of Dm8 depend on R7 activity and Dm8s expression of the histamine chloride channel ORT, indicating that Dm8 receive inhibitory signal from photoreceptors via ORT and convey sign-inverted signal to Tm5c via kainate receptors. In contract to UV preference, color vision in Drosophila is mediated by multiple and partially redundant pathways. To identify color-vision circuits, we developed a novel aversive operant conditioning assay to test animals ability for intensity-independent color discrimination. We successfully trained single flies to discriminate between equiluminant blue or green stimuli. We found that wild-type flies can be trained to avoid either blue or green while mutant lacking functional R7 and R8 photoreceptors can not, indicating that the color entrainment requires the function of the narrow-spectrum photoreceptors R7s and/or R8s. Furthermore, inactivating four types of first-order interneurons, Tm5a/b/c and Tm20, abolishes color learning but inactivating different subsets of these neurons is insufficient to block color learning. In summary, our results suggest that true color vision is mediated by parallel pathways with redundant functions. The apparent redundancy in learned color discrimination is in sharp contrast to innate spectral preference, which is mediated by a single pathway, R7s->Dm8->Tm5c.